Efficient Large-Scale Sweep and Prune Methods with AABB Insertion and Removal

Abstract

We introduce new features for the broad phase algorithm sweep and prune that increase scalability for large virtual reality environments and allow for efficient AABB insertion and removal to support dynamic object creation and destruction. We introduce a novel segmented interval list structure that allows AABB insertion and removal without requiring a full sort of the axes. This algorithm is well-suited to large environments in which many objects are not moving at once. We analyze and test implementations of sweep and prune that include subdivision, batch insertion and removal, and segmented interval lists. Our tests show these techniques provide higher performance than previous sweep and prune methods, and perform better than octrees in temporally coherent environments.